Introduction
aiXplain's Agentic OS is a unified platform for building, deploying, and managing AI agents—from simple automations to complex multi-agent systems. Build with code, deploy instantly, run anywhere; enforce policies in real time and enable continuous self-improvement.
Key Features
- Autonomous agents - Instruction-driven reasoning with multi-step planning, built-in inspectors, and policy enforcement
- Governance & observability - Control at team, agent, and tool levels with usage quotas, rate limits, and execution dashboards
- Code & no-code - Python SDK for developers, Studio UI for builders, OpenAI-compatible API for integration
- Marketplace - 700+ tools and 170+ LLMs—all swappable via unified API with no vendor lock-in
- Data integrations - Vector databases, knowledge graphs, SQL/CSV connectors, and 230+ service integrations (Slack, Google Drive, Salesforce)
- Instant deployment - Deploy agents in seconds as scalable REST or OpenAI-compatible endpoints
- Flexible deployment - Cloud, on-prem, VPC, hybrid, or dedicated infrastructure with containerized portability
Agentic Framework
Core Components
Agents are built from modular components that reason, plan, and act:
- Models - LLMs that interpret instructions, select tools, and adapt execution plans
- Instructions - Natural language prompts guiding agent behavior
- Tools - AI models, Python functions, APIs, MCPs, databases, and other agents accessed through a unified interface
- Memory - Tracks conversation history and maintains context across sessions
- Inspectors - Policy enforcement, content filtering, and compliance validation
Agent
Single-agent system for autonomous, focused tasks using natural language reasoning and tool use.
Use for:
- Q&A with retrieval (Agentic RAG)
- Task automation with tools
- Conversational interfaces
- Multi-turn interactions
Team Agent
Multi-agent system that decomposes complex tasks and coordinates specialized agents with autonomous quality control.
Micro-agents - Built-in coordination layer at every execution loop for planning (Mentalist), validation (Inspector), orchestration (Orchestrator), and response synthesis
Meta-agents - Monitor and optimize agent behavior across multiple runs, autonomously evolving performance over time.
Use for:
- Complex workflows requiring multiple steps
- Tasks needing quality validation
- Cross-functional automation
- Research and analysis pipelines
Pipeline
Deterministic workflows with predefined task sequencing.
Use for:
- Repeatable automation flows
- Strict execution order requirements
- Can be used as tools within agents
Marketplace
Access 700+ ready-to-use AI assets:
- 170+ LLMs - GPT-4, Claude, Gemini, Llama, and more—all swappable via unified API
- Tools & Utilities - Search, OCR, translation, summarization, web scraping
- Integrations - 230+ services via Composio (Slack, Google Drive, Salesforce, etc.)
- Knowledge Tools - Vector databases, graph databases, SQL/CSV connectors
- Custom Assets - Bring your own models, APIs, or Python functions
No vendor lock-in - Use any model or bring your own infrastructure
Deployment & Security
Deployment: Multi-tenant cloud, dedicated cloud, on-premises, hybrid, or private VPC. All agents are containerized and portable.
Security: SOC 2 Type I & II compliant. Data encrypted at rest and in transit. Zero data retention—never stored or used for training. Full API logs and execution tracing. Inspector-based policy enforcement.
Credits & Billing
How Credits Work
- 1 credit = $1 USD
- Example: 1 credit = ~1,500 API calls to GPT-4o Mini
- Buider plan - Pay-as-you-go via Google Pay or credit card (Stripe)
- Team plan — subscription for high-volume usage with optional add-ons
Pricing Model
| Item | Cost |
|---|---|
| Direct model usage | Vendor rates (no markup) |
| Deployed agents/pipelines | Vendor rates + 20% service fee |
| Enterprise subscriptions | Custom pricing with SLAs |
Team Management
- Teams - Share agents, tools, and credits across your organization
- API keys - Team-level or user-level access control
- Usage quotas - Set limits per team, per key, or per model
- Rate limits - Control resource consumption and costs
Observability
Full execution visibility for debugging and auditing:
- Execution trace - Chain of thought, tool calls, and outputs for every step
- Performance metrics - Runtime, API calls, credits consumed, and assets used
- Performance dashboard — live agent KPIs in console.aixplain.com
- Visual agent tracing — graph view of agent flows (coming soon)
Resources
- Security & Compliance - How we handle your data
- Pricing - Detailed cost breakdown
- Discord - Community support
- Enterprise Contact - Custom deployments and SLAs
Next Steps
Ready to build? Check out: Quick Start Tutorial - Build your first agent in 5 minutes